Introduction
According to BCG, in recent years, many companies have focused on boosting their data maturity scores. They report that, in just four years, they were able to grow it from 8% to 15%. Naturally, this vast improvement wouldn’t be possible without investing in building strong, competent data teams, who worked behind the scenes to drive this digital acceleration.
At C&F, we took part in such transformations first-hand, turning our clients from various industries into data-driven and digital industry leaders.
We’re particularly excited to participate in the Gartner Data & Analytics Summit, since one of the themes of the upcoming event will be how leadership can upskill and reskill data teams to boost their business impact. We believe it will be a great opportunity to exchange observations and discuss how AI will drive organizations in their approach to team training.
As we anticipate the start of the conference, here’s our take on the topic.
Organizational models impact the structure of data analytics teams
Data analytics team structures vary across organizations. These differences largely come down to size, the number of projects tackled, as well as operational workflows across the entire business.
For example, it’s common for large companies to have separate data stewards who oversee data governance. Meanwhile, at smaller organizations, their responsibilities will often be assigned to another data team member, for example, a data engineer.
Another factor that may affect what the data team structure looks like is whether the team works on-site, from the same office, or is dispersed across multiple offices or works remotely. In fact, some data teams are built of external consultants, as we will discuss later.
Without further ado, here are some of the most common data team structures out there:
Centralized (top-down) – In this setup, a single team is responsible for managing all of the company’s databases, data collection points, and tech stack. It commonly falls under the broader IT department and, depending on the existing role, could report to the CTO, CIO, or even CAO. Top-down technical teams are popular because they offer the advantage of having a single group of specialists manage data-related technologies for the entire business. That said, the more projects an organization tackles, the less capable such ‘generalist’ teams are at addressing project-specific data challenges. This brings us to the next team setup type.
Decentralized (bottom-up) – As the name gives away, this is an opposing approach, in which data team members are dispersed across various departments. A data analysis expert might, for instance, work alongside the marketing team and help them find technical solutions to goals or challenges whenever the need arises. This could, for instance, be a predictive analysis project that requires coding skills and sourcing data from numerous data points.
As you can see, such an approach can accelerate the work across all company units, because data specialists already understand the context and perspective. The risk here, however, is that each department would start operating in silos. In the worst-case scenario, data specialists could all individually work with their own platforms, making it hard to integrate with databases from other parts of the business.
Hybrid Team – Alternatively, you could integrate data teams within IT and other business units such as sales, customer success, marketing, etc. The idea behind this model is to have data experts closely aligned with both the technical (IT) and functional (business) sides of the organization. All to ensure faster-decision making, by reducing silos and guaranteeing data consistency across teams.
Data Labs – A data lab is a dedicated space or team that focuses mainly on initiatives related to data management. It gives data scientists, analysts, and other data experts an opportunity to experiment with different data models and test new ideas. It allows for a lot of flexibility and agility, which enables organizations to quickly adapt to changes. It’s best suited for businesses that want to have an in-house data team.
Irrespective of the model you go for, it’s worth remembering that for a data team to be successful and thriving, they need to be given room for experimentation. They work best in organizations that are dynamic and innovative. Data teams must feel comfortable to challenge the status quo and to explore new ideas. Only then will businesses be able to truly benefit from having access to analytical minds, who can master data.
Why should you upskill your Data Analytics team?
You might be wondering, ‘is it really necessary to upskill my data teams, and is the time right’. The shower answer is – it definitely is. The U.S. Bureau of Labor Statistics says that jobs related to data science will grow at an unprecedented rate of 36% between now and 2031. This, as you can imagine, will contribute to the war of talent – the demand is high, while the supply, not necessarily.
What does this situation have to do with upskilling?
If skilled data experts are hard to come by, you can invest in tech talent that you already have on board, and encourage them to pick up new skills – you’ll be able to reduce the skill gap.
However, the benefits of upskilling and reskilling your talent go beyond skill gap reduction – most of all, they have a great impact on boosting employee productivity. For example, a data engineer that has recently gone through training that allowed them to acquire new knowledge and skills will be able to find solutions to issues faster, and work more efficiently – and that’s not all.
Growing your tech team skills also brings another benefit – it can drive their job satisfaction levels and encourage them to stay with your organization for longer. This, in particular, relates to upskilling and reskilling initiatives that involve technologies that are set to drive the future of technology.
For example, Python, which powers a lot of groundbreaking AI developments, is an admired technology by 65% of tech specialists, according to StackOverflow’s 2023 developer survey. Helping your team boost skills that are in demand will give them a sense of purpose, and make them more dedicated to the projects they work on.
Naturally, let’s not forget about how reskilling and upskilling benefits your organization. When you keep a close eye on your data teams’ strong skills and provide timely training to close any skills gaps, you’re keeping your company ahead of your competitors. A highly skilled, engaged team means that you can provide the highest quality of services for clients, in a timely manner.
What are the must-have roles and skills for data, analytics & AI?
Let’s now take a look at the roles and skills that organizations should invest in in 2024 and beyond.
Roles:
Data scientists – use both quantitative (statistical, algorithmic, and mining) and qualitative (surveys, etc.) methods to build models that will help businesses find answers to pressing problems, for example, why customers churn. Data scientists often hold a degree in computer science, economics, or statistics. They work towards building the organization’s data infrastructure and providing insights that allow businesses to make better decisions, based on data, not gut feeling. To be able to do that, data scientists turn to predictive analytics to develop models that will help decide on the best course of action for specific situations.
AI/ML developers – are primarily responsible for adding machine learning and AI capabilities, like NLP, image recognition, etc., to different applications. Additionally, they collect and prepare data (either independently or in collaboration with data engineers), which serves as input for training AI data models. Some of the skills that AI and ML developers need include tech expertise to integrate and deploy APIs, as well as the ability to recognize and link data assets, ensure data quality, and integrate data from multiple sources.
Analytics and Business Intelligence Developers – work closely with different business partners offering analytical and tech support to generate business intelligence insights. Their job typically revolves around developing reports, dashboards, and visualizations that help businesses make more informed decisions. To perform this role well, ABI developers need skills like data analysis and modeling, data interpretation and warehousing, programming (especially SQL), etc.
Data engineers – data engineers are responsible for transforming ‘raw’ data and turning it into a format that can be used by other specialists – from both technical and non-technical teams – to run their data analytics projects. Their role is to design a sound data pipeline, which allows them to cater to all data needs within the business. They might work alongside other IT specialists as well as marketing or sales departments, which means that they must also possess communication and team collaboration skills.
Data steward – TechTarget perfectly describes this role by calling it the “liaison between IT and the business side of the organization”. Data stewards are responsible for making sure that the company has enforced – and respects – all applicable data security and privacy policies that relate to data and analytics projects. Part of their role is to monitor the work of other teams to spot and prevent any policy violations. While all companies need to have this function to ensure data compliance, not all have a separate ‘data steward’ role. At smaller businesses, these responsibilities will often fall under the belt of data engineers or data scientists.
Skills:
Now that we’ve discussed the roles in data teams, which skills are we potentially looking to upskill or reskill?
Programming – Programming skills lie at the very core of any team member’s work. While not every data discipline calls for top proficiency in languages and relevant frameworks, anyone running data projects needs to have a certain degree of coding knowledge to run data analysis. For many years now, Python has been at the top of ‘in-demand’ programming skills – in big part, due to its use to build and instruct AI systems. In 2023, it even overtook SQL’s spot as the third most used language in the world. Alongside it, are also popular frameworks and libraries for Python, including NumPy, PyTorch, and TensorFlow.
Other languages data analytics and AI teams use to run projects include C++, Java, and R.
Natural Language Processing (NLP) – NLP has risen to such great popularity, because it’s a technology that gives computers the ability to provide results beyond ‘just’ numbers. It lets machines interact with humans in a question-answer fashion. This ML subset will continue to be a vital skill among data teams because of its use in generative AI. One in three organizations surveyed by McKinsey said they’re already using genAI for their business, with 40% planning to expand its use across the organization. That’s why upskilling or reskilling tech staff in this direction might prove beneficial in the years to come.
Advanced ML and AI – On the one hand, it goes without saying that data teams should continue evolving their ability to build AI/ML technology. On the other though, we believe it’s important to emphasize how this skill can be monetized in the B2B sector. Artificial Intelligence is driving revolutionary changes in industries such as life sciences, and companies need to invest in their own AI technology to remain competitive in the market. All this means that tech teams must have a top level of such AI subsets like deep learning and ML.
Cloud computing – All of the technologies we’ve mentioned so far call for vast amounts of computational power and storage. Given that the IT industry is, simultaneously, expected to close in on zero-emission requirements, this calls for the right solutions. Among others, D&A teams must be proficient in cloud platforms like Google Cloud, Azure, and AWS. More importantly, they must know how to optimize data storage and avoid building a huge infrastructure.
Data visualization – being able to make data-driven decisions requires showing data in a way that is easily understandable, and this calls for data visualization skills. Data experts, especially BI developers, must be proficient in using tools like PowerBI, Tableau, and Python libraries like Matplotlib and Seaborn. All of these are constantly updated, so data experts must stay on top of these changes to be able to effectively communicate their findings.
Ethics in AI and data science – while AI has tremendous potential, the more advanced and widespread it becomes, the more issues related to its ethical use arise. These include avoiding bias and ensuring fairness, data privacy and security, consent and autonomy, etc. All data experts who use AI in their work must be aware of the ethical implications and adapt, in order to responsibly use this technology without posing any threat to society.
Communication and storytelling – data management and analytics isn’t just about mathematics, i.e, calculating numbers. It’s equally important to be able to logically explain and show to people who are not as data-proficient what these numbers actually mean. And this puts good communication and storytelling skills at the front – these will help in translating complex data into actionable insights.
How to upskill and reskill your data, analytics, and AI employees
The World Economic Forum estimates that by 2025 half of all employees will require upskilling or reskilling to remain relevant on the market. Suffice it to say, since it’s just a year away, if you haven’t started building your training strategy yet, the time is now. Here are a few tips on how to get started.
- Map your team’s current skills to identify skills gaps
The first thing you’ll have to do before you start upskilling your employees is running a skill gap analysis. This will help you get a good overview of the skills that your company already has and those that employees can pick up fast.
Consider each data team member individually and then bring all of their skills together in a matrix view.
Create a column with the employee’s name, and separate columns for all the hard skills they possess, like programming language and frameworks, and soft skills like communication and presentation. You can give each skill a score between ‘0’ and ‘4’, where 0 is no ability, and 4 is full professional proficiency. Let’s take a look at an example.
Say you have a couple of data scientists on your team, but you lack BI developers. Since data scientists have strong foundations in programming, statistics, and data analysis, they could probably fairly easily pick up skills like advanced data visualization and SQL that could aid them in performing a BI role.
The more closely related to the current skill set the new skills are that you’d like an employee to learn, the faster they’ll become proficient at them.
- Identify the most in-demand skills to focus on
It’s key to think long-term when you plan the upskilling or reskilling of your tech teams. Given the trends in your industry as well as developments in the AI and data analytics field, what skills will your organization need in the next 2-3 years? What technologies or tools do they need to master? Which skills will they need 5 years from now?
According to research, some of the most in-demand skills for 2024 include SQL, business analysis, AI, software development, and emotional intelligence. Do you already have employees with such skills on board? If not, what can you do to fill the gap?
- Decide how to upskill and reskill individual employees
Once you have a bird’s eye view of your entire team’s competencies and skills gaps, you can start strategizing on how to upskill individual staff. For example, some requirements, like knowledge of AI ethics or regulatory compliance, will be obligatory for the entire data team, meaning, everyone will need to have a good understanding of the laws and requirements. Other reskilling and upskilling needs will be individual.
Bear in mind that people learn in different ways; some prefer courses while others would benefit more from a bespoke mentoring program, or from learning on the job.
Sudhir Khatwani, Founder at The Money Mongers, says that “upskilling and reskilling the people in data and AI isn’t just about throwing courses at them. It’s more personal. Setting up learning paths that match what each person wants is the golden ticket. For example, if someone’s interested in diving deeper into machine learning, we line up the courses, find them a mentor, and give them the opportunity to embark on a project that shows their newly learned skills. It’s like crafting a bespoke suit of skills – fits perfectly and looks great on them.
Khatwani adds that “this way, everyone’s learning something they’re genuinely into, which means they’re more likely to stick with it and exceed it. Plus, it makes the team stronger and more excited about what we’re doing”.
As you can see, training programs must also align with the individual’s own plans, which we discuss next.
- Meet the employee’s career objectives with those of your business
Say that one of your AI developers has built several solutions for the commercial real estate industry while working at a previous company. This could make them a great candidate for joining a team of experts catering to this specific sector, since they already know the specificity of the industry. However, before you narrow their services down, you must make sure that this career development plan aligns with their own objectives.
If you fail to incorporate their perspective, they might consider joining another organization, which aligns more closely with their goals.
Another noteworthy aspect is giving employees the opportunity to acquire special certifications. These should be issued by renowned institutions to further boost the employee’s competencies and work morale. That’s what Lucas Ochoa, CEO and Founder at Automat, has chosen when upskilling his staff.
“Back in 2020 and the years after, the talent shortage in tech was a big problem. I can tell you that it was really hard to find people with skills in AI, data, and other technical areas. So, we decided to upskill our own talented team more about AI and data to keep our company good at these things”, he says.
“What we did was encourage them to get special certifications. I believe getting well-known AI certificates from big names can make our employees more confident in working with data and AI. We sponsored them to go for things like Google’s TensorFlow Developer Certificate or Microsoft’s AI Engineer Certification to make their work profiles better”.
- Set clear and achievable objectives
Once you’re clear on what skills the team is to acquire in the foreseeable future, it’s time to turn this into a realistic roadmap.
It can be relatively easy to estimate if there’s a one-off effort, like having your entire data team get a HIPAA compliance certificate. The deadline in this case would be the date by which everyone will be able to complete training.
In the case of individual skills’ upskilling or reskilling though, agreeing on a timeline can be more difficult. It would be easier to break the roadmap into milestones. For example, a developer who’s looking to move from the software development team to the data scientist team could have a goal of increasing their proficiency in Python by 25% in half a year, or mastering frameworks like PyTorch and TensorFlow, used in deep learning, in a similar time.
Make sure to communicate to your upskilled and reskilled staff that these are all estimated timelines. If a goal you’ve set turns out to be unrealistic, adjust it to ensure you’re not overwhelming your employees and to keep them committed to their skills development.
- Offer real-world experience opportunities
Another way to support your data teams on their path is to find opportunities to use newly-acquired skills in practice. Vlad Khorkhorov, CEO and Co-Founder of WebsitePolicies.com, swears by this approach.
“Our data analytics/AI employees can apply the theories they’ve learned in practical scenarios. These hands-on projects, where they work on actual problems faced by the organization, boost our team’s awareness of how these solutions can address real-world challenges”, he says.
“Moreover, real-world experience helps them develop critical thinking, problem-solving, and decision-making skills, which are all vital in data and AI. Among others, these projects teach them to navigate various data sources, analyze data effectively, and derive actionable insights”.
Another option for offering real-world experiences is by organizing hackathons, and that’s what John Butterworth, Founder and CEO at 10kschools, does. He says: “As manager of the AI and data analytics technical team, I’ve utilized internal hackathons to improve the skills of my team while they tackle actual business challenges. These occasions bring together professionals from various fields, including data scientists, software architects, and AI specialists, in order to promote cross-functional learning. In an effort to optimize the pipeline for processing consumer data, participants in a hackathon experimented with different kinds of technology. We swiftly adopted a machine learning model devised by one group that significantly reduced errors in data processing.
Butterworth adds that “this approach facilitated teamwork and innovation while providing employees with the opportunity to implement and improve their skills in a practical, engaging environment. It proved to be a highly effective method for expediting the acquisition of skills, thereby directly enhancing our operational capacities and fostering an environment that promotes ongoing education and innovation”.
- Revisit your reskilling and upskilling strategies continuously
Last, but not least, treat your data employee development programs as a living organism. Naturally, as we’ve mentioned earlier, look at the numbers to spot any anomalies in skills acquisition. This could include both seeing a team achieve a goal within three months instead of the planned six, or the contrary – an employee or department struggling to reach their milestones by a large percent. Both of these situations will help you plan out training timelines in the future more accurately.
That said, step away from your spreadsheets regularly, i.e., talk to team managers and data and AI team members regularly. Listen to their feedback and answer any questions ongoingly.
Staying engaged in your career development programs is the only way to see long-lasting results for your organization.
Upskilling and reskilling data and AI employees isn’t an option; it’s a necessity
The unprecedented growth of AI technologies means that we’re also looking at an equally unprecedented time in career skills development needs. Companies that want to remain relevant need to monitor their teams’ competencies ongoingly and be aware of each employee’s own career objectives.
We can’t wait to hear what Jorgen Heizenberg has to say about the future of data and AI teams, and how they are set to fit within the wider organization.
